Linux Kernel 5.2.9/5.3.5 MVEBU - NAS326 copying large files with Samba

After long time i updated my NAS326 to Debian Buster and used your last kernel

i try 5.2.9 and 5.3.5

If i copy a lage file via samba from my pc to the nas, the copy process aborts

dmsg show this exeption:

[  456.707542] kswapd0: page allocation failure: order:0, mode:0xa20(GFP_ATOMIC), nodemask=(null),cpuset=/,mems_allowed=0
[  456.719060] CPU: 0 PID: 639 Comm: kswapd0 Not tainted 5.3.5-mvebu-tld-1 #1.0
[  456.726831] Hardware name: Marvell Armada 380/385 (Device Tree)
[  456.733485] [<c01176c4>] (unwind_backtrace) from [<c0112018>] (show_stack+0x10/0x14)
[  456.741961] [<c0112018>] (show_stack) from [<c0a6eb44>] (dump_stack+0x70/0x8c)
[  456.749911] [<c0a6eb44>] (dump_stack) from [<c027993c>] (warn_alloc+0xa0/0x164)
[  456.757947] [<c027993c>] (warn_alloc) from [<c027a774>] (__alloc_pages_nodemask+0xce4/0xf28)
[  456.767115] [<c027a774>] (__alloc_pages_nodemask) from [<c027aab0>] (page_frag_alloc+0x58/0x148)
[  456.776634] [<c027aab0>] (page_frag_alloc) from [<c08d4d48>] (__netdev_alloc_skb+0x104/0x16c)
[  456.785894] [<c08d4d48>] (__netdev_alloc_skb) from [<c0739c20>] (mvneta_poll+0x860/0xb8c)
[  456.794802] [<c0739c20>] (mvneta_poll) from [<c08ecca0>] (net_rx_action+0x170/0x438)
[  456.803273] [<c08ecca0>] (net_rx_action) from [<c0102b84>] (__do_softirq+0x29c/0x3e0)
[  456.811831] [<c0102b84>] (__do_softirq) from [<c012d7c0>] (irq_exit+0x78/0xd0)
[  456.819781] [<c012d7c0>] (irq_exit) from [<c01814c4>] (__handle_domain_irq+0x94/0xe0)
[  456.828342] [<c01814c4>] (__handle_domain_irq) from [<c05da9e4>] (gic_handle_irq+0x58/0x90)
[  456.837423] [<c05da9e4>] (gic_handle_irq) from [<c01021cc>] (__irq_svc+0x6c/0xa8)
[  456.845628] Exception stack(0xdd867d18 to 0xdd867d60)
[  456.851395] 7d00:                                                       dfcf2860 00000cc0
[  456.860300] 7d20: 00000000 00002005 dfcf2864 dfcf2860 dd867e18 df310748 00000001 dd867f14
[  456.869205] 7d40: 00000001 00000000 00000000 dd867d68 c024ad90 c02397a8 400f0013 ffffffff
[  456.878111] [<c01021cc>] (__irq_svc) from [<c02397a8>] (try_to_release_page+0x14/0x68)
[  456.886756] [<c02397a8>] (try_to_release_page) from [<c024ad90>] (shrink_page_list+0x840/0xdac)
[  456.896185] [<c024ad90>] (shrink_page_list) from [<c024b970>] (shrink_inactive_list+0x220/0x420)
[  456.905700] [<c024b970>] (shrink_inactive_list) from [<c024c688>] (shrink_node+0x704/0x88c)
[  456.914779] [<c024c688>] (shrink_node) from [<c024d658>] (kswapd+0x5cc/0x868)
[  456.922640] [<c024d658>] (kswapd) from [<c01458b8>] (kthread+0x130/0x144)
[  456.930152] [<c01458b8>] (kthread) from [<c01010e8>] (ret_from_fork+0x14/0x2c)
[  456.938095] Exception stack(0xdd867fb0 to 0xdd867ff8)
[  456.943861] 7fa0:                                     00000000 00000000 00000000 00000000
[  456.952765] 7fc0: 00000000 00000000 00000000 00000000 00000000 00000000 00000000 00000000
[  456.961669] 7fe0: 00000000 00000000 00000000 00000000 00000013 00000000
[  456.969059] Mem-Info:
[  456.972058] active_anon:3422 inactive_anon:4485 isolated_anon:0
                active_file:19266 inactive_file:81726 isolated_file:32
                unevictable:0 dirty:6499 writeback:3836 unstable:0
                slab_reclaimable:6780 slab_unreclaimable:2435
                mapped:7048 shmem:968 pagetables:301 bounce:0
                free:804 free_pcp:180 free_cma:0
[  457.009743] Node 0 active_anon:13688kB inactive_anon:17940kB active_file:77064kB inactive_file:326904kB unevictable:0kB isolated(anon):0kB isolated(file):128kB mapped:28192kB dirty:25996kB writeback:15344kB shmem:3872kB writeback_tmp:0kB unstable:0kB all_unreclaimable? no
[  457.034619] Normal free:3216kB min:9128kB low:9828kB high:10528kB active_anon:13688kB inactive_anon:17940kB active_file:77064kB inactive_file:326904kB unevictable:0kB writepending:41344kB present:524288kB managed:505092kB mlocked:0kB kernel_stack:760kB pagetables:1204kB bounce:0kB free_pcp:720kB local_pcp:720kB free_cma:0kB
[  457.064101] lowmem_reserve[]: 0 0 0
[  457.068320] Normal: 5*4kB (UME) 3*8kB (UM) 23*16kB (UM) 8*32kB (ME) 4*64kB (UME) 18*128kB (UM) 0*256kB 0*512kB 0*1024kB 0*2048kB 0*4096kB = 3228kB
[  457.082223] 101966 total pagecache pages
[  457.086876] 0 pages in swap cache
[  457.090907] Swap cache stats: add 0, delete 0, find 0/0
[  457.096862] Free swap  = 499708kB
[  457.100884] Total swap = 499708kB
[  457.104913] 131072 pages RAM
[  457.108499] 0 pages HighMem/MovableOnly
[  457.113043] 4799 pages reserved
[  457.116905] mvneta f1070000.ethernet eth0: Can't allocate skb on queue 0
[  457.124331] kswapd0: page allocation failure: order:0, mode:0xa20(GFP_ATOMIC), nodemask=(null),cpuset=/,mems_allowed=0
[  457.135776] CPU: 0 PID: 639 Comm: kswapd0 Not tainted 5.3.5-mvebu-tld-1 #1.0
[  457.143545] Hardware name: Marvell Armada 380/385 (Device Tree)
...

Does anyone have any idea where the problem might be?

EDIT:
I also try out 5.1.7 and 5.1.2



Edited 5 time(s). Last edit at 10/20/2019 05:54AM by bodhi.

Kirsch,

Try increase your swap file size, see if it helps.

How big is the file that you tried to copy?

-bodhi
===========================
Forum Wiki
bodhi's corner



Edited 1 time(s). Last edit at 10/20/2019 05:49AM by bodhi.

Hi Bodhi,

13GB

after ~0,5GB, i get the exeption;

i open "top" in a ssh session
the physical RAM is completely used
the 500MB swap on my flashdrive a untouched

Kirsch,

I think something about your memory management settings that prevented the pages to be evicted.

Are you running systemd?

On sysvinit, I'd set this for server:

cat /etc/sysctl.d/local.conf | grep -i vm

# increase the free page caches for VM
vm.min_free_kbytes = 65536
vm.swappiness = 60

===
Edited: typos.

-bodhi
===========================
Forum Wiki
bodhi's corner



Edited 2 time(s). Last edit at 10/22/2019 10:51PM by bodhi.

> Are you running systemd?
Yes

I add the two lines to my config-file

and now file transfer works fine

thx



Edited 1 time(s). Last edit at 10/21/2019 11:24AM by Kirsch.

Cool!

-bodhi
===========================
Forum Wiki
bodhi's corner